Pro-apoptotic Cleavage Products of Bcl-x L Form Cytochrome c -conducting Pores in Pure Lipid Membranes*

During apoptotic cell death, cells usually release apoptogenic proteins such as cytochrome c from the mitochondrial intermembrane space. If Bcl-2 family proteins induce such release by increasing outer mitochondrial membrane permeability, then the pro-apoptotic, but not anti-apoptotic activity of these proteins should correlate with their permeabilization of membranes to cytochrome c . Here, we tested this hypothesis using pro-survival full-length Bcl-x L and pro-death Bcl-x L cleavage products ( (cid:1) N61Bcl-x L and (cid:1) N76Bcl-x L ). Unlike Bcl-x L , (cid:1) N61Bcl-x L and (cid:1) N76Bcl-x L caused the release of cytochrome c from mitochondria in vivo and in vitro . Recombinant (cid:1) N61Bcl-x L and (cid:1) Rh-DOPE, lissamine rhodamine B-labeled 1,2-dioleoylphosphatidylethanolamine; CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonic acid; PAGE, polyacrylamide gel electrophoresis; zVAD-tmK, benzyloxicarbonyl-Val-Ala-Asp-fluoromethyl ketone. The polyclonal anti-Bcl-x L the monoclonal anti-FLAG antibody the monoclonal anti-cytochrome 65791A and 7H8.2C-12 antibodies The monoclonal anti-Bcl-x L 2A1 antibody LipofectAMINE Protein Expression and Purification— Plasmids encoding glutathione S -transferase (GST)-Bcl-x L fusion proteins were constructed by sub- cloning human Bcl-x L cDNA into pGEX2T (Amersham Pharmacia Bio-tech). A stop codon was inserted after Bcl-x L amino acid 212 to generate GST- (cid:1) C20Bcl-x L . The coding sequence for Bcl-x L amino acids 62–233 and 77–233 plus an initiation codon were amplified by polymerase chain reaction to construct GST- (cid:1) N61Bcl-x L and GST- (cid:1) N76Bcl-x L , respectively. The presence of the deletions was confirmed by DNA se- quencing. All constructs were produced as GST fusion proteins from pGEX vectors using Escherichia coli DH1 (cid:1) as the host strain. A 50-ml overnight culture was used to inoculate 1 liter of Luria-Bertani me-dium, which was further incubated at 37 °C until an A 600 value of 0.4 was achieved. The cells were induced with 0.1 m M isopropyl- (cid:2) were performed in planar lipid bilayer membranes formed by the Muel- ler-Rudin technique across a (cid:3) m diameter as described previously (45). Lipid composition was diphytanoylphosphatidylcholine/ diphytanoylphosphatidylserine (1/1, v/v) or DOPC/DOPE/DOPS (1/1/1, v/v). The solution bathing the membrane contained m M KCl, 0.2 m M m M (pH Membrane thinning was monitored visually and by capacitance measurements. After planar membrane formation, a glass micropipette filled with a solution containing protein was brought into close contact until a gigaseal (resistance (cid:4) 20–100 gigaohms) was formed between the glass in the micropipette tip and the lipid bilayer within Then, a constant potential difference was applied to the patched membrane and the resulting was filtered (5-KHz corner H900 active filter, Frequency Haverhill, MA) and stored (100- (cid:3) s sampling rate). Membrane lifetime experiments were done in a Lucite chamber with a hole of (cid:7) 550 (cid:3) M (Warner Inst., CO). Purified proteins were added to the aque-ous sub-phase, and the solution was stirred for 5 min to ensure good mixing. A software program (BROWSE, available upon request) was modified to apply voltage pulses and facilitate membrane lifetime measurements. analysis of cytosolic fractions. Cross-reactive bands verify equal loading. Similar results were obtained in three independent experiments. c , an analytical gel of purified recombinant Bcl-x L proteins used in this study. Recombinant proteins prepared as described under “Experimental Procedures” were resolved on a 4–20% SDS-PAGE gel and stained with Colloidal blue. Lane 1 , Bcl-x L ; lane 2 , (cid:1) C20Bcl-x L ; lane 3 , (cid:1) N61Bcl-x L ; and lane 4 , (cid:1) N76Bcl-x L . d , recombinant Bcl-x L cleavage fragments induce release of cytochrome c from isolated rat liver mito- chondria into the supernatant ( Sup. ). Mitochondria were treated with recom- binant Bcl-x L , (cid:1) N61Bcl-x L , or (cid:1) N76Bcl-x L , and release of cytochrome c ( cyt c ) was determined by immunoblotting as explained under “Experimental Proce- dures.” Results are representative of four separate experiments.

Proteins of the Bcl-2 family are key regulators of programmed cell death in multicellular organisms. Some members of this family, including Bax, Bak, Bok/Mtd, Bad, Bik/Nbk, Bid, Blk, Bim/Bod, and Hrk promote apoptosis, whereas others, including Bcl-2, Bcl-x L , Bcl-w, Bfl-1/A1, Mcl-1, and Boo/Diva inhibit apoptosis (1). All these proteins share one to four conserved Bcl-2 homology domains (BH) 1 designated BH1, BH2, BH3, and BH4 (1,2). In addition, Bcl-2 family members can possess a C-terminal hydrophobic amino acid sequence that helps localize them to intracellular membranes, primarily the outer mitochondrial membrane (1,2). The activity of Bcl-2 family proteins can be modulated not only at the transcriptional level but also by post-translational modifications (1,3). For example, various cellular proteases have been shown to cleave Bcl-2, Bcl-x L , Bid, Bax, and Bad producing C-terminal fragments with potent pro-apoptotic activity (4 -23). Bcl-x L can be cleaved by caspase 3 after aspartate 61 and 76 and by calpain after alanine 60, converting Bcl-x L from an antiapoptotic factor to a pro-apoptotic factor (5,6,11).
Cumulative evidence indicates that Bcl-2 relatives function, at least in part, by regulating the release of proteins enclosed in the mitochondrial intermembrane space. Current models propose that Bcl-2 family proteins exert this function either by forming pores in mitochondrial membranes themselves, or by modulating endogenous mitochondrial channels through protein-protein interactions (1, 24 -26). Of note, these two mechanisms of action are not necessarily mutually exclusive. The pore-forming function was first proposed based on the structural similarity of Bcl-x L and the pore-forming domain of bacterial toxins such as colicins and the diphtheria toxin (27). More recently, Bid, Bax, and Bcl-2 have shown similar structural patterns (28 -31). Consistent with the structural similarity to bacterial toxins, Bcl-2 family proteins can permeabilize artificial lipid bilayer membranes, with pro-apoptotic members generally showing higher capacity for pore formation than antiapoptotic members (32)(33)(34)(35)(36)(37)(38)(39)(40). Here, we report that pro-apoptotic C-terminal cleavage fragments of Bcl-x L , but not their fulllength anti-apoptotic counterpart, permeabilize both mitochondria and pure lipid bilayer membranes to cytochrome c. lar Probes (Eugene, OR). Sephadex G-10, Sephacryl S-300, and Superdex 200 were purchased from Amersham Pharmacia Biotech (Piscataway, NJ). Caspase 3 (catalog no. 201-038-C005, without CHAPS detergent) and zVAD-fmk were obtained from Alexis (San Diego, CA). The polyclonal anti-Bcl-x L H62 antibody and the monoclonal anti-FLAG antibody were obtained from Santa Cruz Biotechnology (Santa Cruz, CA), and the monoclonal anti-cytochrome c 65791A and 7H8.2C-12 antibodies were from PharMingen (San Diego, CA). The monoclonal anti-Bcl-x L 2A1 antibody was a generous gift from Dr. Larry Boise (University of Miami). LipofectAMINE was obtained from Invitrogen (Carlsbad, CA).
Protein Expression and Purification-Plasmids encoding glutathione S-transferase (GST)-Bcl-x L fusion proteins were constructed by subcloning human Bcl-x L cDNA into pGEX2T (Amersham Pharmacia Biotech). A stop codon was inserted after Bcl-x L amino acid 212 to generate GST-⌬C20Bcl-x L . The coding sequence for Bcl-x L amino acids 62-233 and 77-233 plus an initiation codon were amplified by polymerase chain reaction to construct GST-⌬N61Bcl-x L and GST-⌬N76Bcl-x L , respectively. The presence of the deletions was confirmed by DNA sequencing. All constructs were produced as GST fusion proteins from pGEX vectors using Escherichia coli DH1␣ as the host strain. A 50-ml overnight culture was used to inoculate 1 liter of Luria-Bertani medium, which was further incubated at 37°C until an A 600 value of 0.4 was achieved. The cells were induced with 0.1 mM isopropyl-␤-D-thiogalactopyranoside and incubated at 30°C for an additional 6 h before harvesting by centrifugation. Then, cells were resuspended in a 10-ml phosphate-buffered saline (PBS) solution with 1% Triton X-100 (w/v) and protease inhibitors, and disrupted by sonication. The resulting lysate was centrifuged at 12,000 ϫ g for 15 min at 4°C to pellet cellular debris. Glutathione-agarose beads were added to the supernatant and incubated at 4°C with gentle rotation for 3 h. The beads were washed twice with 50 ml of ice-cold PBS without Triton X-100, packed in a 1-ml column, and washed again with 10 ml of PBS. After incubation with thrombin for 3 h at room temperature, cleaved proteins were eluted with 5 ml of PBS containing 0.5% Nonidet P-40. To 1-ml elution fractions 200 l of prewashed detergent-binding beads (Calbiosorb, Calbiochem, San Diego, CA) were added, followed by incubation for 30 min at 4°C, and removal of the beads by centrifugation. After repeating this process five times, the supernatants were dialyzed against PBS buffer (10-kDa cut-off dialysis membrane), and finally, protein samples were concentrated using a Centricon BIOMAX-10 filter. Purified proteins were characterized by SDS-PAGE in 4 -20% Tris-glycine gels (Invitrogen) followed by Colloidal Blue staining (Pierce, Rockford, IL).
Release of Cytochrome c from Mitochondria-For the immunofluorescence studies, 2-3 ϫ 10 4 BHK cells were seeded on a slide and transiently transfected with 1 g of the appropriate plasmid DNA using LipofectAMINE. 10 h post-transfection, cells were immunostained with sc-807 anti-FLAG and 65791A anti-cytochrome c antibodies, followed by rhodamine-conjugated and fluorescein isothiocyanate (FITC)-conjugated secondary antibodies, respectively (Chemicon, Temecula, CA). Cells were analyzed with a Noran OZ CLSM confocal microscope system with interversion software (Noran Inc., Madison, WI). Cell-fractionation studies were performed as described previously (41). Mitochondria were isolated from livers of male Harlan Sprague-Dawley rats as described by Rickwood et al. (42), followed by two washes in 210 mM mannitol, 70 mM sucrose, 0.5 mM EDTA, and 10 mM HEPES (pH 7.2). Isolated mitochondria (50 g of protein) were incubated with recombinant Bcl-x L protein (1 M) in 50 l of 125 mM KCl, 0.02 mM EDTA, 5 mM sodium succinate, 10 mM HEPES-KOH, pH 7.2, 5 mM Na 2 HPO 4 , 5 M rotenone, for 30 min at 30°C. Reaction mixtures were centrifuged at 14,000 ϫ g for 10 min, mitochondrial membrane pellets corresponding to 25 g of protein and the corresponding volume of supernatants were separated by SDS-PAGE on 4 -20% Tris-glycine gels, and their respective cytochrome c contents were estimated by immunoblotting (7H8.2C-12) using the ECL method (Amersham Pharmacia Biotech).
Liposome Binding Assays-Proteins (100 nM) with or without LUV (300 M lipid) were incubated in buffer A at 37°C for 30 min. Then, 170 l of 73% sucrose (w/v) in buffer A was added to 80 l of the reaction mixture, and the samples were carefully overlaid with 150 l of 40% sucrose in buffer A, and 300 l of 25% sucrose in buffer A. After overnight centrifugation at 215,000 ϫ g, seven 100-l fractions were drawn from the air-fluid interface at the top of the tube. Proteins were precipitated by the chloroform/methanol method and analyzed by immunoblotting using H62 or 2A1 anti-Bcl-x L antibodies. In samples of LUV doped with 0.25 mol% of lissamine rhodamine B-labeled dioleoylphosphatidylethanolamine (Rh-DOPE), the three uppermost fractions of the gradient contained 95% of the fluorescence whereas the three lowermost fractions of the gradient did not show significant fluorescence. Therefore, the first three and the last three fractions collected from the air-fluid interface of the gradient corresponded to the lipid-containing and lipid-free fractions of the gradient, respectively.
Fluorimetric Measurements-Leakage of fluorescent molecules from LUV was continuously monitored in an SLM-2 Aminco-Bowman luminescence spectrometer (Spectronic Instruments), in a thermostatted 1-cm path length cuvette with constant stirring, at 37°C. For ANTS, ex ϭ 355 nm, em ϭ 520 nm (slits ϭ 4 nm); for FITC-dextrans, ex ϭ 490, em ϭ 520 nm (slits ϭ 2 nm). Release of fluorescent markers was quantified on a percentage basis according to the equation, where F f is the measured fluorescence after protein addition, F 0 is the initial fluorescence of the intact LUV suspension, and F 100 is the fluorescence value after complete disruption of vesicle integrity by addition of Triton X-100 (final concentration, 0.2% w/v). Extents of marker release were obtained after fluorescence intensity reached a plateau. Vesicles were stored at 4°C and used within 1-2 days. No spontaneous leakage of entrapped material was observed for vesicles stored at 4°C for at least 1 week. Release of Cytochrome c from LUV-To study cytochrome c release from LUV, two different methods were used. (i) LUV were prepared in buffer A supplemented with 0.1 mM cytochrome c as described above. Then, cytochrome c-containing LUV were incubated with Bcl-x L proteins, 100-l aliquots of treated samples were centrifuged in a 100-kDa cutoff microconcentrator (Millipore Co., Bedford, MA), and finally, cytochrome c contents of the retentate and of the filtrate were analyzed by immunoblotting using 7H8.2C-12 antibody. (ii) LUV were prepared in buffer A supplemented with 0.1 mM cytochrome c and 0.015 mM FD-70 as described above, except that two polycarbonate filters with 0.2-m pores were used for extrusion. The liposomes were incubated with Bcl-x L proteins, 1-ml aliquots of treated samples were loaded onto the Sephacryl S-400 HR column, and the cytochrome c and FD-70 contents of eluted fractions (1 ml) were determined by absorbance and by fluorescence measurements, respectively.
Planar Bilayer Membrane Studies-Single-channel measurements were performed in planar lipid bilayer membranes formed by the Mueller-Rudin technique across a 250-m diameter hole, as described previously (45). Lipid composition was diphytanoylphosphatidylcholine/ diphytanoylphosphatidylserine (1/1, v/v) or DOPC/DOPE/DOPS (1/1/1, v/v). The solution bathing the membrane contained 100 mM KCl, 0.2 mM EDTA, 10 mM HEPES (pH 7.0). Membrane thinning was monitored visually and by capacitance measurements. After planar membrane formation, a glass micropipette filled with a solution containing the protein was brought into close contact until a gigaseal (resistance ϭ 20 -100 gigaohms) was formed between the glass in the micropipette tip and the lipid bilayer within it. Then, a constant potential difference was applied to the patched membrane area, and the resulting current was filtered (5-KHz corner frequency, H900 active filter, Frequency Devices, Haverhill, MA) and stored (100-s sampling rate). Membrane lifetime experiments were done in a Lucite chamber with a hole of ϳ550 M (Warner Inst., Hamden, CO). Purified proteins were added to the aqueous sub-phase, and the solution was stirred for 5 min to ensure good mixing. A software program (BROWSE, available upon request) was modified to apply voltage pulses and facilitate membrane lifetime measurements.

RESULTS
Pro-apoptotic Bcl-x L Cleavage Products Release Cytochrome c from Mitochondria-Because the pro-apoptotic activity of Bcl-2 family proteins appears to rely on their ability to release cytochrome c from mitochondria, we first studied the effect of the pro-apoptotic C-terminal portion of Bcl-x L on cytochrome c distribution in living cells. Expression of ⌬N61Bcl-x L in BHK cells led to a diffuse cytochrome c staining pattern indicating that cytochrome c was no longer confined to the intermembrane space of the mitochondria, but rather it was free in the cytosol (Fig. 1a, filled arrows). In the control neighboring untransfected cells, cytochrome c remained punctuate (open arrows). In agreement with the immunofluorescence studies, subcellular fractionation of BHK cells transfected with ⌬N76Bcl-x L , but not Bcl-x L , showed cytochrome c in the cytosolic fraction (Fig.   1b). These results are similar to those obtained previously using the pro-apoptotic cleaved form of Bcl-2 (41). To investigate whether the release of cytochrome c was a direct or indirect effect of Bcl-x L cleavage products on mitochondria, we tested the ability of purified, recombinant ⌬N61Bcl-x L and ⌬N76Bcl-x L (Fig. 1c) to release cytochrome c from isolated rat liver mitochondria. Incubating isolated mitochondria with recombinant ⌬N61Bcl-x L or ⌬N76Bcl-x L , but not Bcl-x L , led to cytochrome c release (Fig. 1d). Taken together, these data suggest that, similar to other Bcl-2 family proteins with proapoptotic activity (13-16, 20 -23, 46 -51), C-terminal Bcl-x L cleavage fragments act directly on mitochondria to induce cytochrome c release.
Pro-apoptotic Bcl-x L Cleavage Products Possess Intrinsic Pore-forming Activity-To assess whether pro-apoptotic forms of Bcl-x L increase membrane permeability, we studied their effect on pure lipid vesicles loaded with the fluorophore ANTS and its quencher, DPX. ⌬N61Bcl-x L and ⌬N76Bcl-x L induced fast and extensive ANTS release from the liposomes at neutral pH and 1/1000 protein/lipid molar ratios, demonstrating that they possessed endogenous pore-forming activity (Fig. 2a). Anti-apoptotic Bcl-x L caused little ANTS release from LUV, and ⌬C20Bcl-x L , a Bcl-x L construct lacking the C-terminal hydrophobic domain proposed to function as a membrane-anchor domain, had no effect. Of note, ⌬N76Bcl-x L and ⌬N61Bcl-x L did not induce ANTS release through vesicle fragmentation or aggregation, because they caused no significant changes in the static or dynamic light scattering of the LUV suspension (data not shown). To examine whether the different membrane-perturbing effects of these proteins corresponded to their membrane-binding affinities, proteins were incubated with LUV, then free and membrane bound proteins were separated by FIG. 1. Pro-apoptotic Bcl-x L cleavage fragments induce cytochrome c release from mitochondria. a and b, expression of ⌬N61Bcl-x L and ⌬N76Bcl-x L in BHK cells induces translocation of cytochrome c from the mitochondria into the cytoplasm. a, BHK cells transfected with FLAG-tagged ⌬N61Bcl-x L were immunostained with primary antibodies anti-FLAG (left panels) and anti-cytochrome c (right panels), followed by rhodamineconjugated and FITC-conjugated secondary antibodies, respectively. Similar results were obtained in two independent experiments and with ⌬N76Bcl-x L . b, ⌬N76Bcl-x L , but not Bcl-x L , induces release cytochrome c from the mitochondria of BHK cells as determined by fractionation of transfected cells followed by immunoblot analysis of cytosolic fractions. Cross-reactive bands verify equal loading. Similar results were obtained in three independent experiments. c, an analytical gel of purified recombinant Bcl-x L proteins used in this study. Recombinant proteins prepared as described under "Experimental Procedures" were resolved on a 4-20% SDS-PAGE gel and stained with Colloidal blue. equilibrium sedimentation in a discontinuous sucrose gradient. In the absence of LUV, Bcl-x L and its fragments remained in the bottom fractions of the sucrose gradient. However, after incubation with the liposomes, most of the Bcl-x L , ⌬N61Bcl-x L , and ⌬N76Bcl-x L floated to the top fraction of the gradient, demonstrating that they bound to LUV (Fig. 2a, right). ⌬C20Bcl-x L did not bind to LUV at neutral pH but did bind to LUV at low pH. In the course of these experiments we noted that the capacity of Bcl-x L , ⌬N61Bcl-x L , and ⌬N76Bcl-x L to release ANTS from LUV depended on the amount of DOPG present in the liposomes (Fig. 2b, left). This effect was likely due to the higher affinity of Bcl-x L proteins for LUV containing negatively charged lipids (Fig. 2b, right).
Caspase 3 can cleave Bcl-x L to generate ⌬N61Bcl-x L and ⌬N76Bcl-x L (5, 6). Thus, we decided to test whether caspase 3-mediated cleavage of Bcl-x L affects its pore-forming activity. To this aim, Bcl-x L was first allowed to bind LUV containing ANTS/DPX, followed by treatment with caspase 3. Addition of caspase 3 to LUV-associated Bcl-x L increased its pore-forming activity to levels similar to those obtained with ⌬N76Bcl-x L , and this effect was eliminated by the caspase inhibitor zVADtmk (Fig. 3a). Furthermore, the time course of caspase-induced Bcl-x L cleavage correlated with that of ANTS release from liposomes, and zVAD inhibited both processes (Fig. 3b, and data not shown). Finally, similar to the results obtained with the recombinant C-terminal fragments, caspase-cleaved Bcl-x L required the presence of acidic lipids in LUV for efficient binding and pore formation (Fig. 3c). In short, these results indicate that recombinant pro-apoptotic Bcl-x L cleavage products as well as caspase-cleaved Bcl-x L possess endogenous pore-forming activity.
Pro ⌬N61Bcl-x L and ⌬N76Bcl-x L induced release of dextrans of all sizes tested but with different efficiencies (Fig. 4 and Table I). At a constant ⌬N76Bcl-x L concentration, the larger the entrapped dextran, the slower the kinetics and the lower the final extents of dextran release (Fig. 4a). To investigate the dependence of marker release on pro-apoptotic protein concentration, the protein/lipid ratio was varied over a wide range, from 1/10,000 to 1/250 mol/mol (Fig. 4b). ⌬N61Bcl-x L induced half-maximal releases of ANTS, FD-4, FD-10, FD-40, and FD-70 at about 10, 30, 50, 175, and 225 nM, respectively (Fig. 4b). Thus, the larger the encapsulated molecule, the higher the protein concentration required for release. Caspase cleaved Bcl-x L also induced release of high molecular weight dextrans, albeit with lower efficiency compared with ⌬N61Bcl-x L and ⌬N76Bcl-x L (Table I). However, full-length Bcl-x L did not release significant amounts of FITCdextrans from LUV. Prompted by these observations, we decided to test whether the pro-apoptotic Bcl-x L forms could directly release cytochrome c from LUV. To this aim, LUV containing cytochrome c were incubated with Bcl-x L proteins, the treated samples were filtered through a 100-kDa microconcentrator, and the cytochrome c contents of the retentate and the filtrate were detected by immunoblotting. ⌬N61Bcl-x L and ⌬N76Bcl-x L released cytochrome c from LUV in a timeand dose-dependent manner similar to that observed with FD-10 ( Fig. 5, and data not shown). On the contrary, Bcl-x L and ⌬C20Bcl-x L did not cause significant release of cytochrome c. To confirm these findings and to further explore the effect of protein concentration on liposome permeabilization, gel filtration on a Sephacryl 400 HR column was used (Fig. 6). Pure LUV eluted in the void volume of the column, at 12-16 ml (Fig.  6a), whereas free FD-70 and cytochrome c eluted later, at 19 -27 ml and 26 -35 ml, respectively (Fig. 6b). When LUV containing FD-70 and cytochrome c were treated with ⌬N76Bcl-x L at 1/1000 protein/lipid molar ratios, most of the FD-70 eluted together with the liposomes, while most of the cytochrome c did not co-elute with the liposomes (Fig. 6c). Thus, under these conditions ⌬N76Bcl-x L efficiently released cytochrome c but not FD-70. However, at higher ⌬N76Bcl-x L concentrations most of the FD-70 escaped from LUV, similar to cytochrome c (Fig. 6d). Finally, Bcl-x L released neither cytochrome c nor FD-70 (Fig. 6f). In summary, these results demonstrated that, unlike anti-apoptotic full-length Bcl-x L , proapoptotic cleavage products of Bcl-x L formed pores in pure lipid vesicles that allowed passage of large macromolecules, such as cytochrome c and high molecular weight dextrans.    Oligomeric Status of Bcl-x L Proteins-Bax and Bak have been proposed to form cytochrome c-conducting pores through their ability to oligomerize (39,40,(52)(53)(54), and the proapoptotic forms of Bcl-x L may function in a similar manner. To test this hypothesis, we studied the oligomeric status of Bcl-x L proteins by size-exclusion chromatography on Superdex S-200. Bacterially expressed, purified full-length Bcl-x L and ⌬N76Bcl-x L migrated as large complexes; no sign of Bcl-x L monomers (estimated mass, 26,063 Da) or ⌬N76Bcl-x L monomers (estimated mass, 17,774 Da) was detected (Fig. 7a). In contrast, ⌬C20Bcl-x L eluted in a single peak, close to its calculated monomeric molecular mass (24,020 Da), suggesting that the C-terminal hydrophobic domain of Bcl-x L is important for Bcl-x L multimerization. To study the oligomeric status of these proteins after binding to lipid membranes, they were incubated with LUV, followed by solubilization of the membranes with 2% (w/v) CHAPS. Under these conditions, Bcl-x L eluted at molecular masses between ϳ60 and ϳ145 kDa, most consistent with dimers-pentamers of this protein, while ⌬N76Bcl-x L migrated at molecular masses ranging from ϳ35 to ϳ100 kDa, also corresponding to multimers of two to five protein molecules (Fig. 7b).
Therefore, anti-apoptotic and pro-apoptotic forms of Bcl-x L possessed the same oligomeric status in the plane of the bilayer. Hence, oligomerization alone cannot account for the increased pore-forming activity and apoptogenicity of cleaved Bcl-x L .
⌬N61Bcl-x L and ⌬N76Bcl-x L Destabilize Planar Lipid Membranes and Decrease Membrane Line Tension-To gain more insight into the structure of the pores formed by these proteins, electrophysiology of planar phospholipid membranes was used. Soon after adding ⌬N76Bcl-x L , membrane conductance increased by heterogeneous fluctuations in which amplitude increased with time until the membrane became unstable (Fig. 8a). Decreased protein concentration caused similar but delayed electrical responses (data not shown). On the other hand, addition of full-length Bcl-x L induced current fluctuations of ϳ40 -200 pS, but the membranes did not become noticeably unstable (Fig. 8b). To quantify the effect of Bcl-x L proteins on planar lipid bilayer stability, membrane lifetime experiments were performed.
⌬N61Bcl-x L and ⌬N76Bcl-x L , but not Bcl-x L or ⌬C20Bcl-x L , potently diminished planar membrane lifetime (Fig. 8c). By fitting the voltage dependence of membrane lifetime to a theoretical expression for lipidic pore formation (55), line tensions in the absence and presence of ⌬N61Bcl-x L and ⌬N76Bcl-x L were obtained (Fig. 8d). Pro-apoptotic Bcl-x L fragments effectively decreased membrane line tension in a protein-concentration-dependent fashion, reducing its original value by ϳ50% at 15 nM (Fig. 8e). These results are consistent with the hypothesis that pro-apoptotic cleavage fragments of Bcl-x L but not anti-apoptotic full-length Bcl-x L destabilize planar lipid bilayers through reducing membrane line tension. DISCUSSION In this work we investigated the molecular mechanism underlying the pro-apoptotic activity of cleaved Bcl-x L . Our results provide compelling evidence that cleavage of Bcl-x L markedly enhances its pore-forming activity. The C-terminal cleavage fragment of Bcl-x L found in dying cells is sufficient for this activity. The cleavage-induced increase in pore-forming activity likely accounts for the apoptosis-inducing function of cleaved Bcl-x L through the permeabilization of mitochondrial membranes to release apoptogenic intermembrane proteins. Post-transcriptional modifications can affect the function of Bcl-2 family proteins by modulating their subcellular localization (1,3). The pro-apoptotic activity of Bcl-x L elicited by proteolytic cleavage could be due, at least in part, to an increased affinity of the cleaved fragments for mitochondrial membranes, as shown for tBid (13)(14)(15)(16). However, Bcl-x L and its C-terminal cleavage products bound similarly to LUV. Furthermore, in situ cleavage of membrane-bound Bcl-x L increased its poreforming activity to levels similar to those obtained with ⌬N76Bcl-x L . Bcl-x L , but not Bid, contains a hydrophobic Cterminal domain that has been proposed to function as a membrane-anchoring domain (1)(2)(3). Although deletion of this hydrophobic domain greatly impaired the association of Bcl-x L to LUV, efficient binding of full-length Bcl-x L , ⌬N76Bcl-x L , and ⌬N61Bcl-x L to LUV required the presence of anionic lipids in the liposomal membrane. Because Bcl-x L has a biased charge distribution with basic residues being concentrated in one region of the molecule (56), this positively charged region might help protein docking to the membrane surfaces rich in negatively charged lipids. Alternatively, the C terminus of Bcl-x L may lie within the hydrophobic pocket delimited by the BH1, -2, and -3 domains, as recently shown for Bax (30), and anionic lipids may induce a conformational change in Bcl-x L leading to exposure of the hydrophobic C terminus for insertion into membrane. Irrespective of the exact mechanism underlying the association of Bcl-x L with intracellular membranes in vivo, our results suggest that the increased pore-forming activity induced by proteolytic cleavage of Bcl-x L is likely independent of changes in its intracellular distribution.
How do mitochondrially localized Bcl-2 relatives induce release of cytochrome c? In particular, do they act on mitochondrial membranes directly, through their endogenous pore-forming activity, or indirectly, through their ability to interact with other proteins (24 -26)? Our finding, that pro-apoptotic Bcl-x L cleavage fragments efficiently release cytochrome c from pure lipid vesicles, is consistent with the idea that these proteins can, by themselves, form a pathway for cytochrome c release in the outer mitochondrial membrane. Moreover, because the proapoptotic forms of Bcl-x L can release fluorescent dextrans of sizes larger than cytochrome c from LUV, they may also participate in the release of other apoptogenic proteins enclosed in the mitochondrial intermembrane space (26,57,58). However, it should be stressed that our findings do not exclude the possibility that additional cellular components not present in our system may regulate the membrane-perturbing activity of cleaved Bcl-x L in vivo, nor are they necessarily incompatible with other proposed mechanisms of action (59).
The capacity of Bax and Bak to induce permeabilization of pure lipid bilayers and mitochondrial membranes appears to be related to their ability to multimerize (39, 40, 52-54, 60 -62). Similarly, pro-apoptotic Bcl-x L cleavage products may oligomerize as one step in a process of large pore formation in lipid bilayer membranes. In agreement with this idea, gel filtration chromatography studies showed that ⌬N76Bcl-x L formed multimers composed of two to five protein molecules in CHAPS-solubilized LUV. However, Bcl-x L oligomerized in a manner similar to ⌬N76Bcl-x L despite forming pores much smaller than ⌬N76Bcl-x L . Perhaps the oligomers of ⌬N76Bcl-x L penetrate the membrane to a different extent than the Bcl-x L oligomers. Interestingly, a recent study showed that Bcl-x L exists as membrane-integrated large molecular weight complexes in CHAPS-solubilized mi- tochondrial extracts of HeLa cells (61). Although the exact composition of such intramembranous Bcl-x L complexes remains to be determined, this finding raises the possibility that Bcl-x L may also exist in a multimeric form in vivo, at least under certain conditions. Future studies are needed to determine the exact status of endogenous Bcl-x L and cleaved Bcl-x L , as well as their relation to the apoptotic permeabilization of mitochondrial membranes.
Pore formation by Bcl-2 family proteins is likely to be a multistep process coordinated through a set of conformational changes in these molecules. Such conformational changes of Bcl-2 family members may modulate not only protein-protein interactions, but also their effect in the surrounding membrane lipid environment. We have previously proposed that Bax may form pores together with lipid molecules (37). The highly variable membrane conductance changes and decreased membrane stability induced by ⌬N61Bcl-x L and ⌬N76Bcl-x L in planar lipid bilayer membranes suggest that pro-apoptotic Bcl-x L cleavage products may form lipid-containing pores as well.
According to a general model for such pores, the energy E of a pore of radius r can be described as, where ␥ is the line tension of the membrane that creates a barrier against pore opening due to the hydrophobic nature of lipid molecules, and is the surface tension of the membrane that tends to open the pore (55). Because ␥ can originate from the energetic cost to curve lipid molecules in the edge of the pore and because Bax, ⌬N61Bcl-x L , and ⌬N76Bcl-x L reduce ␥ in a concentration-dependent manner (37, and present work), proapoptotic Bcl-2 proteins may form pores through lipid monolayer bending. From this perspective, because the energetic cost to induce lipid monolayer bending is relatively high (63), this process is more likely to proceed through the concerted action of several protein molecules, rather than through changes at the level of isolated protein monomers. In other words, oligomerization of pro-apoptotic proteins may assist in the formation of lipidic pores, as previously proposed for a number of bacterial toxins, membranolytic peptides, and membrane fusion proteins (64 -71).
In summary, our results show that the autonomous poreforming function of cleaved Bcl-x L is sufficient to induce efflux of apoptogenic proteins entrapped within the mitochondrial intermembrane space. The necessity and redundancy of this mechanism for physiological apoptosis, as well as the exact structure of the pore formed by pro-apoptotic proteins in mitochondrial membranes, await further studies.